weather-scraper/scrapers/3Bmeteo.js

const FORMAT = {
    hourly: {
        temperature: {
            type: 'number',
            unit: '°C',
        },
        precipitation: {
            type: 'number',
            unit: 'mm',
        },
        apparentTemperature: {
            type: 'number',
            unit: '°C',
        },
        weatherCode: {
            type: 'string',
        },
    },
    daily: {
        minimumTemperature: {
            type: 'number',
            unit: '°C',
        },
        maximumTemperature: {
            type: 'number',
            unit: '°C',
        },
        minimumApparentTemperature: {
            type: 'number',
            unit: '°C',
        },
        maximumApparentTemperature: {
            type: 'number',
            unit: '°C',
        },
        precipitationSum: {
            type: 'number',
            unit: 'mm',
        },
    },
}

const parseTemp = el => {
    return parseFloat(el.textContent)
}
const parsePrec = el => {
    const text = el.textContent.trim()
    if (text === 'assenti') return 0
    if (text === 'deboli') return 0.05

    try {
        //TODO not 100% sure that this would work as it was never tested
        return parseFloat(text)
    } catch (error) {
        console.log(error)
        return text
    }
}

// const getStartTime = isToday => {
//     if (!isToday) return 0
//
//     const d = new Date()
//     const h = d.getHours()
//
//     if (d.getMinutes() > 30) return h + 1
//     return h
// }
const scrapePage = async (browser, url, isToday) => {
    let result = {}

    const page = await browser.newPage()
    await page.goto(url)

    const tablePrevisioni = await page
        .locator('.table-previsioni-ora')
        .waitHandle()
    const rows = await tablePrevisioni.$$('.row-table.noPad')

    // const startTime = getStartTime(isToday)
    const startTime = isToday ? new Date().getHours() + 1 : 0
    const endTime = 23

    for (let i = startTime; i <= endTime; i++) {
        try {
            const row = rows[rows.length - 1 - (endTime - i)]
            const [rowLeft, rowRight] = await row.$$(':scope > div')

            const codeDiv = (await (await rowLeft.$('.row-table')).$$('div'))[2]
            const rightFields = await (await rowRight.$('.row-table')).$$('div')
            const tempDiv = rightFields[0]
            const precDiv = rightFields[1]
            const appTempDiv = rightFields[4]

            const weatherCode = await codeDiv.evaluate(el =>
                el.textContent.trim(),
            )
            const temperature = await (
                await tempDiv.$('span')
            ).evaluate(parseTemp)
            const precipitation = await precDiv.evaluate(parsePrec)
            const apparentTemperature = await (
                await appTempDiv.$('span')
            ).evaluate(parseTemp)

            result[i] = {
                temperature,
                precipitation,
                apparentTemperature,
                weatherCode,
            }
        } catch (error) {
            result[i] = null
            console.log(error)
        }
    }

    await page.close()

    return result
}

const getDaySummary = day => {
    let minimumTemperature = Number.MAX_VALUE
    let maximumTemperature = Number.MIN_VALUE
    let minimumApparentTemperature = Number.MAX_VALUE
    let maximumApparentTemperature = Number.MIN_VALUE
    let precipitationSum = 0
    for (const h in day) {
        minimumTemperature = Math.min(minimumTemperature, day[h]?.temperature)
        maximumTemperature = Math.max(maximumTemperature, day[h]?.temperature)
        minimumApparentTemperature = Math.min(
            minimumApparentTemperature,
            day[h]?.apparentTemperature,
        )
        maximumApparentTemperature = Math.max(
            maximumApparentTemperature,
            day[h]?.apparentTemperature,
        )
        precipitationSum += day[h]?.precipitation
    }
    return {
        minimumTemperature,
        maximumTemperature,
        minimumApparentTemperature,
        maximumApparentTemperature,
        precipitationSum,
    }
}

export default async browser => {
    const url = text => `https://www.3bmeteo.com/meteo/pisa/${text}`

    const results = await Promise.all([
        scrapePage(browser, url(''), true),
        scrapePage(browser, url('1'), false),
        scrapePage(browser, url('2'), false),
        scrapePage(browser, url('3'), false),
        scrapePage(browser, url('4'), false),
        scrapePage(browser, url('5'), false),
        scrapePage(browser, url('6'), false),
    ])

    return {
        format: FORMAT,
        today: results[0],
        tomorrow: results[1],
        dayAfterTomorrow: results[2],
        week: { ...results.map(getDaySummary) },
    }
}